This is my submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass.
🌱 What I Built
What if technology could help us spend less time on our screens and more time enjoying the outdoors?
That is the idea behind Touch Grass Outdoor Guide, a web application designed to help people plan simple outdoor activities and reconnect with nature.
The project brings together weather information, nearby green spaces, AI-assisted outing plans, nature-based activities, and progress tracking in one place.
Instead of encouraging people to spend more time browsing, the goal is to make planning easier so they can put their phones away and enjoy the real world.
The main features include:
- 🌤️ Weather and Air Quality: View current weather conditions, air-quality information, and a seven-day forecast when the connected services are available.
- 🗺️ Nearby Parks and Gardens: Explore nearby green spaces using map-based discovery.
- 🥾 Smart Outdoor Planner: Create a personalized outing plan based on preferences such as activity, duration, mood, and fitness level.
- 🎒 Packing Checklist: Get suggested items to take along on an outdoor activity.
- 🛡️ Safety Reminders: View basic reminders to help prepare for an outing.
- 📵 Screen-Free Challenge: Encourage yourself to spend less time on your phone during outdoor activities.
- 🌿 Nature Bingo: Turn outdoor exploration into a simple activity by noticing and discovering things around you.
- 🏆 My Progress: Track outing history, points, achievements, and personal goals in the browser.
The Outdoor Score is a simple planning indicator based on available information. It is not an official weather warning or a guarantee that a location is safe.
The central idea is simple:
Plan less. Go outside more. Notice the world around you.
🎬 Project Demo
🌐 Live Website:
https://touch-grass-outdoor-guide.onrender.com/
🎥 Video Demo:
https://www.image2url.com/r2/default/videos/1791722093710-c0de9100-a1d8-428e-8a81-a45cc1f09998.mp4
The video demonstrates the application's interface and main features. Some features depend on external services, so the walkthrough should not be treated as proof that every live API is currently working.
💻 Source Code
GitHub Repository:
https://github.com/tanmaydave2007-netizen/touch-grass-outdoor-guide
The repository contains the application source code and setup instructions for running the project locally.
🛠️ How I Built It
I built Touch Grass Outdoor Guide using a Python-based backend and a web frontend.
The main technologies include:
- Python and Flask: Handle server-side functionality and application routes.
- HTML, CSS, and JavaScript: Build the user interface and interactive features.
- Chart.js: Display forecast and progress visualizations.
- Leaflet: Display interactive maps.
- OpenStreetMap: Provide community-maintained map and green-space data.
- Open-Meteo: Provide geocoding, weather, and air-quality data.
- OpenRouter: Connect the application to a selected AI model for outing-plan generation.
- Browser Local Storage: Keep progress and outing information in the user's browser.
Open-Weight AI Integration
The application supports connecting an AI model through OpenRouter's chat-completions API.
One intended model option is Gemma 4 26B A4B Instruct, with the model ID:
google/gemma-4-26b-a4b-it:free
Model availability may change, and the exact model should be confirmed from a successful response before claiming that it generated a particular plan.
The AI planner is designed to turn user preferences and available context into a practical outdoor plan with suggested steps, a checklist, and reminders.
When no API key is configured, the application can display a clearly labelled demo plan. That fallback is a local template, not an AI-generated response.
This implementation uses hosted inference through OpenRouter. The model weights are not downloaded or run locally by this application.
🔓 Why Does Open Innovation Matter?
For me, open innovation means having more flexibility to learn, experiment, and build useful applications using AI technologies.
1. Flexibility to Experiment
An open-weight model gives developers more opportunities to explore different model choices, adjust prompts, and adapt a solution to a specific use case.
For this project, the goal is to use AI for practical outdoor planning rather than just generating general-purpose text.
2. More Control Over the Application
A model-connected application can be designed around a specific task, with structured outputs such as activities, checklists, and reminders.
The application can also be adapted as different models and tools become available, subject to their technical requirements and licences.
3. Experiment Without Running Large Models Locally
I chose hosted inference because it allows me to experiment without downloading large model weights or requiring a powerful GPU on my own computer.
However, hosted inference still requires an internet connection and may be subject to provider availability, usage limits, or costs.
4. Honest About Privacy and Limitations
Open-weight AI does not automatically mean that an application is offline or completely private.
In this project, prompts sent through OpenRouter may be processed by OpenRouter and the selected model provider. Users should avoid including sensitive personal information.
I want to understand not just how to connect an AI model, but also how the model is accessed, what information is sent, and what limitations come with that approach.
🌍 How It Fits the Touch Grass Theme
The purpose of this application is to make technology a starting point for an outdoor experience, rather than the main activity.
The intended workflow is straightforward:
- Check the available weather information.
- Explore nearby parks or green spaces.
- Create a plan for a short outdoor activity.
- Use the checklist to prepare for the outing.
- Step outside, observe nature, and spend less time looking at the screen.
- Record progress and plan another outdoor activity.
The app combines planning tools with small nature-based activities to make getting outside feel more approachable.
🚧 Current Limitations and Next Steps
Touch Grass Outdoor Guide is a working prototype, and some parts still need testing and improvement.
The public deployment is available, but the live weather endpoint is currently returning an error in my tests. I am investigating the connection to the external weather services.
My next steps are to:
- Fix and verify the public weather and air-quality requests.
- Test the complete AI-planning workflow using the exact selected model.
- Improve error handling when external services are unavailable.
- Test nearby-park discovery with different locations.
- Verify that the application works consistently across desktop and mobile browsers.
- Improve the setup instructions so other developers can run the project more easily.
- Take the application on a real outdoor walk and document what works and what needs improvement.
I have not included any unverified outdoor field-test results or claimed that all public features are working reliably.
💡 What I Learned
Building this project has helped me understand how different technologies work together in a web application.
I have explored how Python and Flask can connect frontend interactions to backend services, how external APIs can provide weather and map data, and how an AI model can be integrated into a practical planning workflow.
I also learned that connecting an AI model is only one part of building a useful application. Reliability, privacy, user experience, error handling, and honest communication about limitations matter too.
The most important lesson is that AI should help people do something meaningful in the real world—not simply keep them engaged with another screen.
🌿 Final Thought
Less scrolling. More exploring. More time in nature.
I built Touch Grass Outdoor Guide as a step toward making outdoor activities easier to plan and encouraging people to reconnect with the world around them.
Feedback, suggestions, and contributions are welcome!
Top comments (0)